CN108830271A - A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks - Google Patents
A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks Download PDFInfo
- Publication number
- CN108830271A CN108830271A CN201810607174.2A CN201810607174A CN108830271A CN 108830271 A CN108830271 A CN 108830271A CN 201810607174 A CN201810607174 A CN 201810607174A CN 108830271 A CN108830271 A CN 108830271A
- Authority
- CN
- China
- Prior art keywords
- reading
- recognition
- convolutional neural
- neural networks
- digital displaying
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
Digital displaying meter Recognition of Reading method the invention discloses one of pattern-recognition and field of artificial intelligence based on convolutional neural networks, including data acquisition, data processing, the building of depth network model and trained, meter reading identification and etc..The present invention is realized a kind of high-precision digital displaying meter reading automatic identifying method, is had the characteristics that recognition accuracy height, strong real-time, have preferable practical value by the learning training process based on big data Instrument image.
Description
Technical field
The present invention relates to pattern-recognitions and field of artificial intelligence, in particular to a kind of be based on convolutional Neural
The digital displaying meter Recognition of Reading method of network.
Background technique
Being automatically identified in various measurements and monitoring system for meter reading has a wide range of applications.Such as water, electricity and gas heat
In meter reading charging application, need periodically to read meter reading.In monitoring system, it is also desirable to read instrument periodically or in real time and read
Number, to realize monitoring and control to system.Currently, the acquisition of meter reading is mainly the following mode:
(1)In some application fields, meter reading is also mainly by the way of manually reading, such as water, electricity and gas dsc data is artificial
Meter reading.This mode is time-consuming and laborious, is also unfavorable for the automation of system.
(2)The automatic acquisition of meter reading can be read by measuring instrumentss digital improvement with directly acquiring digitlization
Number.However, this mode needs biggish cost input.Such as digital improvement is carried out to existing water meter, it needs to dismantle pipeline
And digital water gauge is replaced, investment is larger, also makes troubles to user.In addition, stem-winder has digitlization table in some fields
It is difficult to the advantage substituted.Such as mechanical water meter has at low cost, measurement is accurate, without advantages such as power supplys.
(3)Reading automatic identification based on computer vision shoots Instrument image by camera, utilizes computer vision
Technology automatic identification meter reading.This method has plug and play, without being transformed original measuring instrumentss, it is low in cost the features such as.
Currently, digital displaying meter based on computer vision reads automatic identifying method, method for distinguishing is mainly known using individual character,
It is single number that image segmentation, which will be read, then utilizes traditional classifier to each number(Such as support vector machines)It carries out
Identification.The deficiency of this method is to be easy the interference by picture noise during Character segmentation, generate and accidentally divide, identification is accurate
Rate is not also high.
Drawbacks described above is worth solving.
Summary of the invention
In order to overcome the shortcomings of that existing technology, the present invention provide a kind of digital displaying meter reading based on convolutional neural networks
Recognition methods.Technical solution of the present invention is as described below:
A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks, which is characterized in that include the following steps:
S1:Data acquisition:It include the Instrument image of reading area using picture pick-up device shooting;
S2:Data processing:By artificial notation methods, meter reading value is labeled;
S3:The building of depth network model and training:Instrument image and corresponding labeled data are input to depth network model to carry out
It trains, parameter learning is carried out using residual error passback algorithm in training process;
S4:Meter reading identification:Instrument image is inputted, depth network model provides corresponding meter reading recognition result and identification
Confidence level.
According to the present invention of above scheme, which is characterized in that in the step S1, when the Instrument image of shooting, instrument
Dial reading part face camera lens, it is placed in the middle and occupy image area 2/3rds or more.
According to the present invention of above scheme, which is characterized in that in the step S2, marked content is meter reading, is read
", " is used to separate between each of several numbers.
Further, for just in the reading position of carry, X.5 annotation formatting is.
According to the present invention of above scheme, which is characterized in that in the step S3, specifically include step:
S31:Construct deep neural network model;
S32:The setting of network model training parameter;
S33:Random initializtion is carried out to model parameter, the training of deep neural network model is then carried out, obtains identification model.
Further, in the step S31, deep neural network model includes characteristic extracting module, contextual information
Fusion Module and categorization module:Wherein, the characteristic extracting module extracts digital feature information from meter reading image, passes through
Convolutional neural networks obtain advanced features figure;The contextual information Fusion Module carries out context according to advanced features information
The fusion of information;The categorization module contextual information module enhances digital advanced features, and carries out to reading
Classification and prediction.
Further, in the step S32, network model training parameter include the number of iterations, optimizer, learning rate,
Learning rate more new strategy and weight attenuation coefficient.
Further, the number of iterations is 1000000, and the optimizer uses ADADELTA method, the study
Rate is 1.0, and the learning rate more new strategy uses changeless learning rate, and the weight attenuation coefficient is 0.0005.
Further, in the step S33, training process using residual error passback algorithm carry out parameter learning, by from
The last layer of model starts to calculate transmitting residual error, and successively to front transfer, to be updated to model parameter, to reach net
The purpose of network training.
According to the present invention of above scheme, which is characterized in that in the step S4, specifically include following steps:
S41:Meter reading image is inputted into network, using training obtained identification model and parameter, to reading carry out prediction and
Identification, obtains recognition result;
S42:Recognition result is returned, and arithmetic average is carried out to the recognition confidence of each place reading, obtains recognition confidence.
According to the present invention of above scheme, the beneficial effect is that:
1, compared with traditional artificial meter reading, the present invention can be realized the meter reading identification of full automation, save a large amount of manpowers
Material resources, while in terms of real-time, high-volume, it has a clear superiority.
2, compared with existing reading automatic identifying method based on computer vision, the present invention is had the following advantages that:
(1)The present invention uses depth network architecture, avoids and carries out explicit Character segmentation, feature extraction and character classification
Process.Model automatic learning data distribution, model can be suitable for the feature of image of real scene, identify from mass data
Precision is higher.
(2)The present invention is obtained more robust filter, is coped with by residual error back propagation algorithm, adjust automatically convolution kernel
The complicated practical application scene such as fuzzy, perspective transform, light conversion.
Detailed description of the invention
Fig. 1 is overall flow block diagram of the invention.
Fig. 2 is the structural schematic diagram of depth network model of the present invention.
Fig. 3 is the schematic diagram of digital displaying meter Recognition of Reading example of the present invention.
Specific embodiment
With reference to the accompanying drawing and the present invention is further described in embodiment:
As shown in Figure 1, a kind of digital displaying meter Recognition of Reading method based on convolutional neural networks, including at data acquisition, data
The processes such as reason, the building of depth network model and the identification of trained and meter reading.
One, data acquisition:Instrument dial plate image is shot using picture pick-up devices such as mobile phone, specialized hardwares.
In shooting process, instrument dial plate reading portion divides face camera lens, it is placed in the middle and occupy image area 2/3rds or more,
Instrument dial plate allows to tilt to a certain degree, but reading area needs to shoot completely.Captured Instrument image should cover different rule
The instrument to be detected of lattice, Instrument image quantity are no less than 100000.
Two, data processing:The image of shooting is labeled, marked content is meter reading, each of reading number
Between use ", " to separate, marked content is without including any coordinate information.
Wherein, for just in the reading position of carry, annotation formatting be X.5, such as certain reading position value between 7 and 8,
It then needs to be labeled as " 7.5 ".Every shared 0-9 i.e. 10 digital states, in addition carry status, then every co-exists in 20 classes mark
State.
Three, the building of depth network model and training:Instrument image and corresponding labeled data are input to depth network model
It is trained, parameter learning is carried out using residual error passback algorithm in training process.Specifically include following steps:
1, deep neural network model is constructed
As shown in Fig. 2, constructed deep neural network model includes characteristic extracting module module, contextual information Fusion Module
With categorization module module.Wherein, characteristic extracting module extracts digital feature information from meter reading image, passes through convolutional Neural
Network obtains advanced features figure;Contextual information Fusion Module carries out the fusion of contextual information according to advanced features information;Point
Generic module contextual information module enhances digital advanced features, and reading is classified and predicted.
(1)Characteristic extracting module structure is:
Network layer | Concrete operations | Characteristic pattern size |
Input layer | - | 3×40×144 |
Convolutional layer | Nuclear volume 64, convolution kernel 3 × 3, step-length 1 × 1 mend side | 64×40×144 |
Non-linear layer | - | 64×40×144 |
Pond layer | Pond core 2 × 2, step-length 2 × 2 | 64×20×72 |
Convolutional layer | Nuclear volume 64, convolution kernel 3 × 3, step-length 1 × 1 mend side | 64×20×72 |
Non-linear layer | - | 64×20×72 |
Pond layer | Pond core 2 × 2, step-length 2 × 2 | 64×10×36 |
Convolutional layer | Nuclear volume 128, convolution kernel 3 × 3, step-length 1 × 1 mend side | 128×10×36 |
Non-linear layer | - | 128×10×36 |
Pond layer | Pond core 2 × 1, step-length 2 × 1 | 128×5×36 |
Convolutional layer | Nuclear volume 128, convolution kernel 2 × 2, step-length 1 × 1 | 128×4×35 |
Non-linear layer | - | 128×4×35 |
Convolutional layer | Nuclear volume 256, convolution kernel 3 × 3, step-length 1 × 1 mend side | 256×4×35 |
Non-linear layer | - | 256×4×35 |
Pond layer | Pond core 2 × 2, step-length 2 × 2 | 256×2×18 |
Convolutional layer | Nuclear volume 512, convolution kernel 2 × 2, step-length 1 × 1 | 512×1×17 |
Normalize layer | - | 512×1×17 |
Non-linear layer | - | 512×1×17 |
As shown above, in characteristic extracting module, the welt operation of convolutional layer is, in pasting up and down for former characteristic pattern
Upper circle pixel, pixel value 0;Non-linear layer uses ReLU activation primitive;Pond layer is using maximum pond mode;Normalization
Each characteristic pattern is normalized to Gauss normal distribution by layer.
(2)Contextual information Fusion Module structure is:
Network layer | Concrete operations | Characteristic pattern size |
Long short-term memory layer | Nodal point number 128 | 128×1×17 |
Long short-term memory layer | Nodal point number 256 | 256×1×17 |
As shown above, in contextual information Fusion Module, long short-term memory layer is LSTM model structure.
(3)Used categorization module structure is:
Network layer | Concrete operations | Characteristic pattern size |
Full articulamentum | Nodal point number 256 | 256×1×17 |
Full articulamentum | Nodal point number 256 | 256×1×17 |
Full articulamentum | Nodal point number 21 | 21×1×17 |
As shown above, categorization module uses the pre- geodesic structure with 17 positions, and prediction result is carried out CTC
(Connectionist Temporal Classification)Decoding, obtains the recognition result of meter reading.
Above-mentioned characteristic extracting module, contextual information Fusion Module and categorization module has used depth network model,
Neural net layer in table is sequential connection form, updates the parameter in neural network using residual error passback algorithm.Wherein:
The input of characteristic extracting module is meter reading image, is exported as advanced features figure, and merges mould as contextual information
The input of block;The output of contextual information Fusion Module is the fusion feature figure of contextual information, and as the defeated of categorization module
Enter;The output of categorization module is the prediction result of 17 positions, and carries out CTC(Connectionist Temporal
Classification)Decoding.3 modules use supervised learning method, by training, study obtain digital picture feature and
The mapping relations of label.
2, the setting of training parameter:
The number of iterations:1000000;
Optimizer:Using ADADELTA method;
Learning rate:1.0;
Learning rate more new strategy:Using changeless learning rate;
Weight decay(Weight attenuation coefficient):0.0005.
3, random initializtion is carried out to model parameter, then carries out the training of deep neural network model, obtain identification mould
Type.
Training process carries out parameter learning using residual error passback algorithm, is transmitted by calculating since the last layer of model
Residual error, and successively to front transfer, to be updated to model parameter, to achieve the purpose that trained network.
Four, meter reading identifies
Meter reading identification process specifically includes:(1)One meter reading image is inputted into network, the identification obtained using training
Model and parameter, predict logarithm and are identified, recognition result is obtained;(2)Recognition result is returned to, and to each place reading
Recognition confidence carries out arithmetic average, obtains recognition confidence.
As shown in figure 3, which show the recognition results and confidence level of digital displaying meter reading.Wherein, left column is recognition result,
It include Recognition of Reading result, recognition confidence, the right side is classified as the image of meter reading.
The present invention uses global recognition method, without explicitly being divided to number in reading, identity with higher
Energy.In actual test, method of the invention can achieve 99.0% recognition accuracy, and recognition speed is every meter reading figure
As being no more than 30 milliseconds, it can satisfy the needs of practical application.
One aspect of the present invention is effectively utilized the feature learning ability and classification performance of depth network model, while can be more
Effectively to find the data characteristics being hidden in a large amount of Instrument image data, thus realize a kind of high-precision meter reading from
Dynamic recognition methods.The present invention has preferable practical application value, can be widely applied to be related to the field of meter reading identification,
In the automatic identification of the various metering meter readings of such as water, electricity, gas, heat.
It should be understood that for those of ordinary skills, it can be modified or changed according to the above description,
And all these modifications and variations should all belong to the protection domain of appended claims of the present invention.
Illustrative description has been carried out to the invention patent above in conjunction with attached drawing, it is clear that the realization of the invention patent not by
The limitation of aforesaid way, if the method concept of the invention patent and the various improvement of technical solution progress are used, or without
It improves and the conception and technical scheme of the invention patent is directly applied into other occasions, be within the scope of the invention.
Claims (9)
1. a kind of digital displaying meter Recognition of Reading method based on convolutional neural networks, which is characterized in that include the following steps:
S1:Data acquisition:It include the Instrument image of reading area using picture pick-up device shooting;
S2:Data processing:By artificial notation methods, meter reading value is labeled;
S3:The building of depth network model and training:Instrument image and corresponding labeled data are input to depth network model to carry out
Training;
S4:Meter reading identification:Instrument image is inputted, depth network model provides corresponding meter reading recognition result and identification
Confidence level.
2. the digital displaying meter Recognition of Reading method according to claim 1 based on convolutional neural networks, which is characterized in that
In the step S1, when the Instrument image of shooting, instrument dial plate reading portion divides face camera lens, placed in the middle and occupy the three of image area
/ bis- or more.
3. the digital displaying meter Recognition of Reading method according to claim 1 based on convolutional neural networks, which is characterized in that
In the step S2, marked content is meter reading, uses ", " to separate between each of reading number.
4. the digital displaying meter Recognition of Reading method according to claim 1 based on convolutional neural networks, which is characterized in that
In the step S3, step is specifically included:
S31:Construct deep neural network model;
S32:The setting of network model training parameter;
S33:Random initializtion is carried out to model parameter, the training of deep neural network model is then carried out, obtains identification model.
5. the digital displaying meter Recognition of Reading method according to claim 4 based on convolutional neural networks, which is characterized in that
In the step S31, deep neural network model includes characteristic extracting module, contextual information Fusion Module and categorization module:
Wherein, the characteristic extracting module extracts digital feature information from meter reading image, obtains height by convolutional neural networks
Grade characteristic pattern;The contextual information Fusion Module carries out the fusion of contextual information according to advanced features information;The classification
Module contextual information module enhances digital advanced features, and reading is classified and predicted.
6. the digital displaying meter Recognition of Reading method according to claim 4 based on convolutional neural networks, which is characterized in that
In the step S32, network model training parameter includes the number of iterations, optimizer, learning rate, learning rate more new strategy and power
Weight attenuation coefficient.
7. the digital displaying meter Recognition of Reading method according to claim 6 based on convolutional neural networks, which is characterized in that institute
Stating the number of iterations is 1000000, and the optimizer uses ADADELTA method, and the learning rate is 1.0, and the learning rate updates
Strategy uses changeless learning rate, and the weight attenuation coefficient is 0.0005.
8. the digital displaying meter Recognition of Reading method according to claim 4 based on convolutional neural networks, which is characterized in that
In the step S33, training process carries out parameter learning using residual error passback algorithm, by counting since the last layer of model
Transmitting residual error is calculated, and successively to front transfer, to be updated to model parameter, to achieve the purpose that network training.
9. the digital displaying meter Recognition of Reading method according to claim 1 based on convolutional neural networks, which is characterized in that
In the step S4, following steps are specifically included:
S41:Meter reading image is inputted into network, using training obtained identification model and parameter, to reading carry out prediction and
Identification, obtains recognition result;
S42:Recognition result is returned, and arithmetic average is carried out to the recognition confidence of each place reading, obtains recognition confidence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810607174.2A CN108830271A (en) | 2018-06-13 | 2018-06-13 | A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810607174.2A CN108830271A (en) | 2018-06-13 | 2018-06-13 | A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108830271A true CN108830271A (en) | 2018-11-16 |
Family
ID=64145112
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810607174.2A Pending CN108830271A (en) | 2018-06-13 | 2018-06-13 | A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108830271A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902751A (en) * | 2019-03-04 | 2019-06-18 | 福州大学 | A kind of dial digital character identifying method merging convolutional neural networks and half-word template matching |
CN110175648A (en) * | 2019-05-28 | 2019-08-27 | 东莞德福得精密五金制品有限公司 | The information communication method of Noninvasive is carried out to equipment using artificial intelligence cloud computing |
CN110490195A (en) * | 2019-08-07 | 2019-11-22 | 桂林电子科技大学 | A kind of water meter dial plate Recognition of Reading method |
CN111027456A (en) * | 2019-12-06 | 2020-04-17 | 四川杰森机电有限公司 | Mechanical water meter reading identification method based on image identification |
CN111024145A (en) * | 2019-12-25 | 2020-04-17 | 北京航天计量测试技术研究所 | Method and device for calibrating handheld digital meter |
CN111666930A (en) * | 2019-03-07 | 2020-09-15 | 深圳海青智盈科技有限公司 | Meter reading method, device, system and storage medium |
CN111783785A (en) * | 2020-07-02 | 2020-10-16 | 上海海事大学 | Water meter identification system and method |
CN111950396A (en) * | 2020-07-27 | 2020-11-17 | 江苏大学 | Instrument reading neural network identification method |
CN113159172A (en) * | 2021-04-20 | 2021-07-23 | 上海济辰水数字科技有限公司 | Intelligent water meter image positioning training method, intelligent water meter identification system and method |
CN113159170A (en) * | 2021-04-20 | 2021-07-23 | 上海济辰水数字科技有限公司 | Intelligent water meter identification system and method based on deep learning |
CN113221959A (en) * | 2021-04-20 | 2021-08-06 | 上海济辰水数字科技有限公司 | Intelligent water meter image recognition training method, intelligent water meter recognition system and intelligent water meter recognition method |
CN113743405A (en) * | 2021-09-07 | 2021-12-03 | 南方电网数字电网研究院有限公司 | Intelligent meter reading method and device for electric energy meter |
CN115439849A (en) * | 2022-09-30 | 2022-12-06 | 杭州电子科技大学 | Instrument digital identification method and system based on dynamic multi-strategy GAN network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105654130A (en) * | 2015-12-30 | 2016-06-08 | 成都数联铭品科技有限公司 | Recurrent neural network-based complex image character sequence recognition system |
CN105678293A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex image and text sequence identification method based on CNN-RNN |
CN107085723A (en) * | 2017-03-27 | 2017-08-22 | 新智认知数据服务有限公司 | A kind of characters on license plate global recognition method based on deep learning model |
CN107133616A (en) * | 2017-04-02 | 2017-09-05 | 南京汇川图像视觉技术有限公司 | A kind of non-division character locating and recognition methods based on deep learning |
-
2018
- 2018-06-13 CN CN201810607174.2A patent/CN108830271A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105654130A (en) * | 2015-12-30 | 2016-06-08 | 成都数联铭品科技有限公司 | Recurrent neural network-based complex image character sequence recognition system |
CN105678293A (en) * | 2015-12-30 | 2016-06-15 | 成都数联铭品科技有限公司 | Complex image and text sequence identification method based on CNN-RNN |
CN107085723A (en) * | 2017-03-27 | 2017-08-22 | 新智认知数据服务有限公司 | A kind of characters on license plate global recognition method based on deep learning model |
CN107133616A (en) * | 2017-04-02 | 2017-09-05 | 南京汇川图像视觉技术有限公司 | A kind of non-division character locating and recognition methods based on deep learning |
Non-Patent Citations (4)
Title |
---|
B. SHI等: ""An End-to-End Trainable Neural Network for Image-Based Sequence Recognition and Its Application to Scene Text Recognition"", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
C. WIGINGTON等: ""Data Augmentation for Recognition of Handwritten Words and Lines Using a CNN-LSTM Network"", 《2017 14TH IAPR INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION (ICDAR)》 * |
J. PUIGCERVER: ""Are Multidimensional Recurrent Layers Really Necessary for Handwritten Text Recognition?"", 《2017 14TH IAPR INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION (ICDAR)》 * |
蔡梦倩等: ""基于全卷积网络的数字仪表字符识别方法"", 《现代计算机(专业版)》 * |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902751B (en) * | 2019-03-04 | 2022-07-08 | 福州大学 | Dial digital character recognition method integrating convolution neural network and half-word template matching |
CN109902751A (en) * | 2019-03-04 | 2019-06-18 | 福州大学 | A kind of dial digital character identifying method merging convolutional neural networks and half-word template matching |
CN111666930A (en) * | 2019-03-07 | 2020-09-15 | 深圳海青智盈科技有限公司 | Meter reading method, device, system and storage medium |
CN110175648A (en) * | 2019-05-28 | 2019-08-27 | 东莞德福得精密五金制品有限公司 | The information communication method of Noninvasive is carried out to equipment using artificial intelligence cloud computing |
CN110175648B (en) * | 2019-05-28 | 2024-01-05 | 东莞德福得精密五金制品有限公司 | Non-invasive information communication method for equipment by applying artificial intelligent cloud computing |
CN110490195A (en) * | 2019-08-07 | 2019-11-22 | 桂林电子科技大学 | A kind of water meter dial plate Recognition of Reading method |
CN111027456A (en) * | 2019-12-06 | 2020-04-17 | 四川杰森机电有限公司 | Mechanical water meter reading identification method based on image identification |
CN111027456B (en) * | 2019-12-06 | 2023-06-20 | 四川杰森机电有限公司 | Mechanical water meter reading identification method based on image identification |
CN111024145A (en) * | 2019-12-25 | 2020-04-17 | 北京航天计量测试技术研究所 | Method and device for calibrating handheld digital meter |
CN111783785A (en) * | 2020-07-02 | 2020-10-16 | 上海海事大学 | Water meter identification system and method |
CN111783785B (en) * | 2020-07-02 | 2024-04-09 | 上海海事大学 | Water meter identification system and identification method |
CN111950396A (en) * | 2020-07-27 | 2020-11-17 | 江苏大学 | Instrument reading neural network identification method |
CN111950396B (en) * | 2020-07-27 | 2024-05-14 | 江苏大学 | Meter reading neural network identification method |
CN113159170A (en) * | 2021-04-20 | 2021-07-23 | 上海济辰水数字科技有限公司 | Intelligent water meter identification system and method based on deep learning |
CN113221959A (en) * | 2021-04-20 | 2021-08-06 | 上海济辰水数字科技有限公司 | Intelligent water meter image recognition training method, intelligent water meter recognition system and intelligent water meter recognition method |
CN113159172A (en) * | 2021-04-20 | 2021-07-23 | 上海济辰水数字科技有限公司 | Intelligent water meter image positioning training method, intelligent water meter identification system and method |
CN113743405A (en) * | 2021-09-07 | 2021-12-03 | 南方电网数字电网研究院有限公司 | Intelligent meter reading method and device for electric energy meter |
CN115439849A (en) * | 2022-09-30 | 2022-12-06 | 杭州电子科技大学 | Instrument digital identification method and system based on dynamic multi-strategy GAN network |
CN115439849B (en) * | 2022-09-30 | 2023-09-08 | 杭州电子科技大学 | Instrument digital identification method and system based on dynamic multi-strategy GAN network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108830271A (en) | A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks | |
CN103324937B (en) | The method and apparatus of label target | |
CN108805070A (en) | A kind of deep learning pedestrian detection method based on built-in terminal | |
CN110956412B (en) | Flood dynamic assessment method, device, medium and equipment based on real-scene model | |
CN113538391A (en) | Photovoltaic defect detection method based on Yolov4 and thermal infrared image | |
CN108921203A (en) | A kind of detection and recognition methods of pointer-type water meter | |
CN110287942A (en) | Training method, age estimation method and the corresponding device of age estimation model | |
CN111353352A (en) | Abnormal behavior detection method and device | |
CN111950396B (en) | Meter reading neural network identification method | |
CN102645173A (en) | Multi-vision-based bridge three-dimensional deformation monitoring method | |
CN111476827A (en) | Target tracking method, system, electronic device and storage medium | |
CN115775085B (en) | Digital twinning-based smart city management method and system | |
CN104616002A (en) | Facial recognition equipment used for judging age groups | |
CN108961308A (en) | A kind of residual error depth characteristic method for tracking target of drift detection | |
CN109165309A (en) | Negative training sample acquisition method, device and model training method, device | |
CN110852243A (en) | Improved YOLOv 3-based road intersection detection method and device | |
CN111079773A (en) | Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network | |
CN109840497A (en) | A kind of pointer-type water meter reading detection method based on deep learning | |
CN113706579A (en) | Prawn multi-target tracking system and method based on industrial culture | |
CN112613438A (en) | Portable online citrus yield measuring instrument | |
CN114861761A (en) | Loop detection method based on twin network characteristics and geometric verification | |
CN111652168B (en) | Group detection method, device, equipment and storage medium based on artificial intelligence | |
CN112529003A (en) | Instrument panel digital identification method based on fast-RCNN | |
CN115760990A (en) | Identification and positioning method of pineapple pistil, electronic equipment and storage medium | |
Zhang et al. | Research on Binocular Stereo Vision Ranging Based on Improved YOLOv5s |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181116 |